Abstract: In order to solve the time-consuming problem of data acquisition of the whole wavefield, a compressed sensing algorithm was introduced to represent the wavefield sparsely. Due to the complex propagation characteristics of guided waves in composite materials, the application of compressed sensing in composite plates becomes difficult. Hence, by considering the anisotropic wavenumber characteristics of the guided waves propagating in the composite plate, a wavenumber library under different angles in the calculated frequency band was constructed for wavefield reconstruction. In the stage of damage analysis, a composite material damage scattering wavefield separation technology without reference signal technology was proposed to accurately remove the incident wavefield and improve the accuracy of damage location. Simulation and experimental results show that the proposed approach allows a reduction of the measurement locations required for accurate signal recovery to less than 90% of the original sampling grid and the single damage location error is less than 2/3 of the minimum wavelength. In addition, the double-damage experimental results show that the proposed method can effectively locate the double-damage, and related results can provide a theoretical and methodological basis for the practical application of composite damage detection.